Accuracy of diabetes diagnosis in health insurance claims data in Taiwan

Abstract

Background and Purpose: There are limited data from Taiwan about die accuracy of National Health Insurance (NHI) claims data. This study assessed the accuracy of NHI claims data for diabetes and its associated factors. Methods: Insurance claims data for patients with a diagnosis of diabetes were extracted from the records of the Bureau of National Health Insurance, including detailed files of the outpatient, emergency, inpatient and pharmacy treatment records from January 1, 2000 to December 31, 2000. A stratified, 2-staged, probability proportional to size and equal probability method was used to sample 9000 diabetes patients. The accuracy of the diabetes diagnosis was assessed based on patient responses to questionaire items. Subjects with negative or uncertain questionaire answers who had hypoglycemic agents in pharmacy claims data were also classified as diabetic. Results: A total of 1350 questionnaires were returned and an accurate diagnosis was verified from data in 1007 (74.6%) of these subjects. Univariate analysis showed that level of accreditation of the hospital, age, gender, follow-up department, type of complication, number of outpatient visits, emergent visit, as well as hospitalization were significant factors associated with an accurate diagnosis of diabetes. Multivariate logistic regression analysis revealed that number of outpatient visits, hospitalization, age, and the level of accreditation of the hospital were significant independent factors. The odds ratio of an accurate diagnosis increased with the number of outpatient visits and hospitalization. The probability of accurate diagnosis of diabetes among patients with ? 4 outpatient visits was 99.16 times greater than that of patients with ? 1 outpatient visit. The probability of accurate diagnosis in patients with A 1 hospitalization was 5.26 times that of patients who had not been hospitalized. Conclusions: This study found that the accuracy of diabetes diagnosis in NHI claims data in Taiwan was 74.6%. Further attention to the association of inaccurate claims in cases with infrequent outpatient visits, young age and those attending non-accredited hospitals is needed in order to address the efficiency of diagnosis and surveillance of diabetes in Taiwan.